Cells within a tumor sample are known to be heterogeneous, due to the contamination from non-malignant tissue or the presence of multiple sub-clones, each carrying different somatic mutations. The majority of somatic mutations identified to date are clustered (mutational hotspots) in the functional sites of a few "cancer genes" with key roles in cell signaling pathways of proliferation and survival. Somatic mutations in cancer genes modify their oncogenic potential or affect sensitivity to therapy. Currently available assays that are able to detect rare somatic mutations are not comprehensive. They are usually focused on a few commonly mutated loci, and not implemented in clinical setting due to cost or technical reasons. Therefore, a current technological need exists for an assay that can reliably detect and accurately measure the prevalence of multiple somatic mutations present only in a fraction of the cells in a heterogeneous tumor. Such an assay would facilitate translational research to study the selection of tumor sub-clones during disease progression and treatment. Additionally the assay could be used by clinicians to improve tumor characterization and selection of therapy choices during clinical trials. We propose to leverage the emerging technology of targeted high-throughput sequencing to develop a cost- effective assay capable of detecting somatic mutations that are present in e1% of tumor cells. Specifically we will perform ultra-deep targeted sequencing (UDT-Seq) of ~100 kb in each tumor assaying 518 mutational hotspots located in 46 cancer genes. The selected mutational hotspots cover ~87% of all entries in the COSMIC database. We will develop a streamlined sample preparation in collaboration with RainDance Technologies to ensure a straightforward implementation in the clinic. This sample preparation integrates the targeting PCR and the library preparation in one step using chimeric PCR primers. The amplified targeted hotspots (200bp long) will be thus directly sequenced on the Illumina Genome Analyzer (GAII) at a very high coverage (~20,000x). We will then precisely model the sequencing error using calibration samples to filter true mutations from the sequencing noise. Our specific aims are: 1) To calibrate the UDT-Seq assay by analyzing both pooled DNA samples containing precise ratios of known SNPs and DNA samples spiked with low amounts of mutated DNA from cancer cells. For this, we will develop a statistical sequencing error model to detect rare mutations in deep sequence. 2) To evaluate the accuracy of the UDT- Seq assay to detect rare somatic mutations in both frozen and formalin fixed paraffin embedded solid tumors. If the quantitative milestones set for this pilot phase of the UDT-Seq assay development are met, we will apply for R33 funding to further develop and make this assay broadly available to clinical oncologists for their own translational research through a CLIA laboratory. PUBLIC HEALTH RELEVANCE: We propose to develop an assay to analyze somatic mutations in rare subclones of solid-tumors. This assay will feature microfluidic-based sample preparation method and high throughput DNA sequencing. It would be the most comprehensive assay to identify rare somatic mutational hotspots in a clinical setting and clinical oncologists will potentially use this assay to identify new biomarkers for disease progression or predictive of drug response, for a more personalized treatment of cancer.